doc-skills-md
Pass
Audited by Gen Agent Trust Hub on Mar 7, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill performs analysis of external project files to generate documentation, which presents a surface for indirect prompt injection.
- Ingestion points: Multiple project files including SKILLS.md, CLAUDE.md, AGENTS.md, package.json, go.mod, pyproject.toml, Dockerfile, and README.md.
- Boundary markers: Absent; there are no specific delimiters or instructions to ignore embedded commands within the analyzed data.
- Capability inventory: File system read operations (cat), local JSON processing (python3), and file system write (updating SKILLS.md).
- Sanitization: Absent; the skill directly interpolates information found in project files into its plugin recommendation logic.
- [COMMAND_EXECUTION]: The skill executes standard read commands (
cat) to access project metadata and the agent's local plugin configuration file (~/.claude/plugins/known_marketplaces.json). - [COMMAND_EXECUTION]: The skill uses
python3 -cfor localized JSON parsing of the marketplace catalog file.
Audit Metadata